On Tue, Sep 29, 2015 at 11:48 PM, Koen Kooi firstname.lastname@example.org wrote:
Op 30 sep. 2015, om 00:37 heeft Rob Herring email@example.com het volgende geschreven:
On Tue, Sep 29, 2015 at 9:42 AM, Ivan T. Ivanov firstname.lastname@example.org wrote:
On Tue, 2015-09-29 at 12:32 +0100, Daniel Thompson wrote:
On 29/09/15 12:03, Grant Likely wrote:
>>> PS >>> Does anyone have a canned command to deliberately corrupt the > boot-and-rootfs so I can get the board to jump into fastboot (so I can > try a snapshot) without having to take the sensors board of the top so I > can press Vol- ;-) ? > > You should be able to use the Sonic Screwdriver 96boardsctl utility to > reset the board. It twiddles the reset line on the LS connector, which > is the same signal as Vol- (at least, so I'm told)
Ah... Reset == Vol-Down. I didn't think if that (Esla: that would
explain why the reset button on the Sonic Screwdriver didn't do what we were expecting).
Anyhow the trick works.
That said, rather frustratingly the trick only works when I can
physically un/plug the power to the main board (if I use the 'screwdriver to simulate a long press off SW2 the board doesn't revisit fastboot properly).
I think we need some help from our Qualcomm friends here on a reset sequence.
Wouldn't do any harm ;-)
The 'screwdriver has access to the power and reset button signals, and I can make them be twiddle in whatever order is required.
Does perhaps Vol- need to be held down before pressing power?
I've tried a fair few combinations.
Pressing and holding SW2 (aka screwdriver power pin) for 15 seconds causes the board to reset itself although, unlike a typical android phone, the board starts again automatically without any need for further presses of SW2. However I haven't (yet?) found any way to make SW4 influence the resulting reboot.
It looks like there is no user-space daemon, which handle SW2 KEY_POWER event and power-off gracefully the system.
IIRC, handling KEY_POWER will work with a GUI running, but for headless you would need acpid running.
No need for acpid If your SoC matches this udev rule: https://github.com/systemd/systemd/blob/master/src/login/70-power-switch.rul... , systemd will catch the event and do a shutdown.
I should have known systemd has consumed this function as well.